How to Understand The Central Role of Exadata to Oracle’s Cloud Strategy
Last Updated on March 21, 2021 by Shaun Snapp
- Exadata is central to Oracle’s cloud strategy.
- Learn why this is a severe problem as Oracle is betting on proprietary designs to defeat open source.
This article will cover the major modalities of computer hardware.
Our References for This Article
If you want to see our references for this article and other related Brightwork articles, see this link.
Lack of Financial Bias Notice: The vast majority of content available on the Internet about Oracle is marketing fiddle-faddle published by Oracle, Oracle partners, or media entities paid by Oracle to run their marketing on the media website. Each one of these entities tries to hide its financial bias from readers. The article below is very different.
- First, it is published by a research entity.
- Second, no one paid for this article to be written, and it is not pretending to inform you while being rigged to sell you software or consulting services. Unlike nearly every other article you will find from Google on this topic, it has had no input from any company's marketing or sales department.
There are four different hardware types/modalities. These are:
- On-premises proprietary servers (which tend to be purchased from major brands like HP, Dell, Lenovo, and Cisco)
- Cloud commodity servers (which are typically run by cloud service entities and customized by those entities)
The appliance is a combination of hardware and software to run a specific type of processing. One of the most prominent examples of this is Exadata by Oracle. Exadata provides a critical case study of how proprietary software and hardware are combined in one hardware modality.
The Central Role of Exadata to Oracle’s Cloud Strategy
If we look at Mark Hurd’s comments, we can see how Exadata is central to Oracle’s cloud strategy.
“We try not to get into this capital expenditure discussion. It’s an interesting thesis that whoever has the most capex wins,” Hurd said in response to a question from Fortune at a Boston event on Tuesday. “If I have two-times faster computers, I don’t need as many data centers. If I can speed up the database, maybe I need one fourth as may data centers. I can go on and on about how tech drives this.
Oracle has said it runs its data centers on Oracle Exadata servers, which are turbocharged machines that differ fundamentally from the bare-bones servers that other public cloud providers deploy by the hundreds of thousands in what is called a scale-out model.”
Oracle uses Exadata machines in their cloud to make the competitive argument against AWS/GCP/Azure. These machines are an inferior fit for absolutely anything other than running the Oracle database. Oracle intends to take its proprietary software and hardware model and compete against cloud service providers that use open-source software and hardware. Because of this, Oracle will lose. Proprietary models only work when the on-premises model can be applied. However, cloud competition allows for a far more open competition. It is somewhat surprising that Oracle has yet to figure this out.
- If Exadata were so cost-effective, as proposed by Mark Hurd, AWS/GCP would purchase them.
- Oracle cannot be more efficient than AWS/GCP by utilizing Oracle hardware/software because Oracle hardware/software is not cost-competitive. Oracle is for customers to purchase by brand. AWS is moving off of Oracle.
- Every time Oracle executives commented about infrastructure and competed with AWS/GCP, it makes no sense and ends up being inaccurate or misrepresenting around the cloud in general.
The problem with Exadata machines is that they’re designed for a single tenant like a bank running on a private LAN. In a multi-tenant public cloud network, you need the most granular level of individual resource control. That’s not what appliances like Exadata are designed to do. This makes adding features extremely complicated, if not impossible.
Oracle’s Misleading Use of the Term Cloud
Something to be aware of is that when Oracle uses the term “cloud,” they only mean this in the most deceptive way that one could imagine. The cloud is based on specific principles. The fact is that most of what SAP and Oracle offer customers is a faux cloud, as we covered in the article How to Understand Oracle as Faux Cloud Versus AWS.
Cloud is frequently diluted to mean the service is merely off-premises. Let us take this opportunity to clarify what the cloud is and what features are necessary for an offering to be considered a cloud.
The following eight conditions must be met for a service to be the cloud.
- One codebase
- No customization
- The vendor provides the hosting (i.e., the vendor provides and maintains all infrastructure for the application)
- Flexible cancellation
- Published and transparent pricing
- Using a cloud salesforce
- Using self-guided demo systems
- Vendor-provided Software maintenance
Oracle is opposed to nearly every item on this list and does not offer flexible terms or transparent pricing. Therefore, when Oracle uses the term cloud, they mean to copy the marketing cache or cloud without providing any of the cloud’s benefits.
Exalogic is the lesser-known compute box that runs the WebLogic application server and Oracle’s middleware and Oracle’s applications (previously called Fusion, but now called Cloud, Oracle ERP Cloud, Oracle Procurement Cloud, etc..). According to iDatalabs, 711 customers use Exalogic. Exalogic competes with the IBM AS400, HP Servers, and Unisys. With less than 1% of the market, it is a high niche offering.
Exalogic is covered in the following quotation:
“Mark Benioff, founder of Salesforce.com, presumes that any appliance principally lacks scalability for the end-user compared with the infrastructure, supplied as service, and notes that the Exalogic approach is actually a rollback to the obsolete mainframe computer concept. Also, commentators have expressed concerns about the appropriateness of placing the word “elastic” in the name, because, despite the ability to load balance, there are obvious computing limits of the box, and those limits cannot be transcended like they should be in a true elastic environment; the same criticism applies to all solutions designed for private cloud computing, in particular, it applies to EMC Corporation and Hewlett-Packard products. However, any computing environment is a collection of servers, and since many Exalogic machines can be combined, it is not limited to the single box capacity, which may be considered as merely a building block.”
Disagreements on Exadata and Oracle’s Cloud Strategy
Thomas Kurian, the previous head of Oracle Cloud (who now works at Google Cloud), spotted the problem with this direction. If Oracle customers migrate to Oracle cloud at scale, Oracle will be forced to move away from their own hardware because it is vastly more cost-effective to use Intel commodity servers rather than Exadata. These issues with Exadata are central to why Thomas Kurian left Oracle.
The first generation Oracle cloud (OCI Classic) was based on Nimbula technology. Nimbula was founded in late 2008 by Chris Pinkham and Willem Van Biljon, who developed the Amazon Elastic Compute Cloud (EC2) and was acquired by Oracle in 2013. Nimbula Director lets you do things like control access to local and external cloud resources with a policy-based authorization system supporting multi-tenancy, hands-off automated installation on bare metal, automated cluster expansion, and Linux and Windows support (VMs). OCI ran on Intel servers, and the first wave of OCI customers was almost entirely lifted and shift from Oracle applications. OCI had minimal adoption within Oracle’s customer base for many reasons, but it moved forward despite all the problems.
Then Larry Ellison decided that the generation two Oracle cloud is going to run on Exadata machines. The idea was that over 50% of Oracle software sales in the past several years were bundled with engineered systems like Exadata and other appliances, not standalone. Also, Oracle hardware and storage sales have been dropping steadily by 5-10% for the past several years. Oracle shrunk the lower-priced server business that it purchased from Sun to focus on the higher-end servers like Exadata. (We discussed earlier how both Oracle and HPE have moved upmarket in servers) This means that Oracle now really only sells high-end servers.
Smaller vendors can only dream of locking in their customers the way that Oracle routinely does. Oracle has the most advanced “value extraction” teams that work both before the sale and after the sale, making the true price paid for Oracle unknowable.
The Lock-In of Exadata
With Exadata, the lock-in is complete. The machine is 100% Oracle hardware (Sun) and software. The issue is that the world is moving toward connected networks using inter-operable and distributed systems. Using any proprietary closed stack is a sure way to get left out and marginalized.
Where Does Exadata Fit?
Exadata is designed to run Oracle software with highly predictable usage requirements on customer premises.
Exadata is a poor fit if:
1. The business requires workload variability (seasonality, user growth, data source variety).
2. The business needs to use non-Oracle data or applications.
3. The business needs the agility to quickly provision capacity up and down or have fine-grained control over resources in a hosted environment.
When Oracle resources discuss Exadata, the lack of flexibility is left out of the equation. For such a device to have a place in the future, it has to be so good at a specific task that it has no competition. This is true of mainframes, for which no other commodity server has been found to compete well, banking applications, ATM processing, data ingestion, or similar. Exadata does not have very much to offer customers that commodity servers cannot effectively do. Furthermore, Exadata requires a dedication to the Oracle database, a database that customers should be making efforts to migrate away from (where possible) due to costs, audit risk, and the availability of excellent open-source alternatives.
Oracle’s focus on Exadata shows the overall DB centricity of Oracle’s thought patterns. They believe that they can compete and differentiate on the basis (in part) of Exadata for cloud — which is a high-cost query optimized appliance for all the different processing types offered by the hyperscale cloud vendors. Everything for Oracle seems to come back to the database.
Oracle is mistaken here.
The database works within and will increasingly work within the context of the overall infrastructure. Bringing out Exadata for cloud will not bring them up to par with AWS/GCP/Azure.
To compete with AWS, Microsoft, Google, and IBM, Oracle needs a differentiator. An answer to “Why Oracle Cloud?” That will never be price or functionality because Oracle’s cost structures are at least an order of magnitude higher than AWS, and Oracle cloud is at least ten years behind AWS in capability.
So, what’s left?
The database, high-performance computing, Exadata on bare-metal.
Is this working?
It does not appear to be because Oracle’s revenue grew 1.64% over the past five years. Oracle’s cloud strategy is to differentiate on high performance both in database and hardware. This is where both bare metal and Exadata come into the picture high-performance offerings for the most demanding workloads. The problem, of course, is that this is a niche market dominated by IBM and offers little if any growth potential. Secondly, Intel-based hardware has come a long way and offers tremendous performance capabilities at a fraction of the price. Exadata runs on Sun hardware and with the Solaris operating system. The market has standardized on Intel/Linux platform for many years already. Not only is Sun/Solaris unpopular, but it also has minimal support in crucial areas like the hypervisor, containerization/virtualization management providers.
Furthermore, Oracle has been disinvesting in both Sun and Solaris through successive layoffs. This long-term trend in the decline of Sun hardware and Solaris is why Sun was vulnerable to acquisition in the first place. Sun was in play to be acquired by HP and others before eventually gobbled up by Oracle. At the time, it was considered an unusual move by Oracle as Sun’s hardware margins had continually eroded, and it had not found very much success with obtaining revenues from its open-source offerings. IBM was beating Sun in Sun’s core market. What looks likely in the future is Exadata and Exalogic being ported to run Linux.
Oracle offers proprietary hardware and cost when both the server OS is open source, and the hardware is being designed to open-source specifications. This makes this part of the cloud challenging to scale without the flexibility and fine-grained control available on the Intel/Linux ecosystem.
Finally, cloud mega-vendors like AWS, GCP, and Azure now offer bare metal in the cloud and even on-premises hardware (outposts).
AWS and Google now offer high-performance computing both in cloud and on-premises. Microsoft offers items at lower price points and has been beating Oracle in database sales since 2015. Larry Ellison had a clear vision for Sun when the acquisition was made, combining Sun’s proprietary hardware with Oracle’s proprietary software to build a lock-in appliance. The problem is that the industry is moving in the opposite direction from this. This is in some way apparent in Oracle’s marketing of Exadata, which cloud washes Exadata and Exalogic. Oracle is attempting to sell its appliances by telling customers they are something that they are not.